5
On the Nature and Measurement of
Fiscal Illusion: A Survey
Wallace E. Oates
Wallace E. Oates lectures at the Departmenr of Economics and Bureau of Business
and Economic Research, U niversiry of Maryland. This chapter was written during
sabbatical leave at Resources for the Future. I am grateful to RFF and to the
Sloan Foundation for their support of this work. I am also indebted to Dennis
Mueller, Paul Porrney, and Robert Schwab for some most helpful commenrs on
an earlier drafto
It is both a pleasure and an honour to contribute to this volume in recognition
of the valuable and wide-ranging contributions of Russell Mathews to the
study of fiscal strucrure and behaviour. Under Mathews's careful guidance,
the Centre for Research on .Federal Financial Relations at the Australian
National U niversiry has provided a steady stream of published research
on federal finance that occupies a central place in the field. Scholars throughout the world draw regularly on Mathews's and the Centre's work both
for its substantive findings and for the rich bodies of comparative institutional information and data that~ they provide. All of us in the field
are deeply in his debt.
For my contribution, I have chosen the very tantalising, but empirically
the notion that the systematic
elusive, phenomenon of 'fiscal illusion'
misperception of key fiscal parameters may significantly distort fiscal choices
by the electorate.
Various elementS of the tax structure, for example ~ay be largely hidden
so that voters do not perceive the entire cost of providing certain public
services. There is now a small literature on this issue: itS historic rootS
lie in Puviani (1903) and other Italian writers; and it has received imperus
moreorecently from James Buchanan (1967).
-
.
The most recent work on fiscal illUsionis largely empirical in charaaer:
it seeks to find particular manifestations of fiscal illusion in existing fiscal
65
-
-
--
.0
---
Taxation and Fiscal Federalism
66
structUre and choices. A central theme of my paper will be that this literatUre
has been less successful than it might appear in establishing the presence
and importance of various forms of fiscal illusion. For reasons that I will
try to make clear, the detection and measurement of fiscal illusion is a
difficult enterprise. In several instances where stUdies claim to find empirical
support for a particular form of fiscal illusion, I believe that there are
alternative, and more compelling explanations for the findings. In other
cases, the evidence simply is not consistent with the illusion hypothesis.
This is not to say that elements of fiscal illusion are nOt present in fiscal
I suspece they are - but rather that the existing empirical
systems
literatUre has not as yet made a persuasive case for their existence and
importance.
The chapter begins with a general consideration of the natUre of fiscal
illusion and then tUrns to empirical efforts to substantiate various illusion
hypOtheses. Five forms (or sources) of fiscal illusion have received attention
in the literatUre:
-
1. complexity of the tax structUre;
2.
renter illusion with respect to property taxation;
3.
income elasticity of the tax structUre;
4.
debt illusion;
5. the flypaper effect.
I shall summarise and evaluate the work on each of them in turn.l But
first we need a brief overview of fiscal illusion and of the general empirical
approach to testing illusion hypotheses.
The Nature
of Fiscal Illusion
Almost 30 years ago, Anthony Downs (1957) argued convincingly that
the representative voter is likely to have highly imperfect information
on which to base his decisions on public-sector activities. The costs of
acquiring information imply, of course, less than perfect knowledge for
private-sector choices as well. But as Downs pointed out, the incentives
to obtain" information on the benefits and costs of government programs
are much less than for private activities. Since an individual ballot is unlikely
to have a discernible effect on public-sector outcomes, there is little reason
for the vOter to invest substantial resources in learning about the relative
costs and benefits of the various alternatives in the government sector.
The individual determines his own pattern of purchases and consumption
of private goods so that some investment in the acquisition of information
is worthwhile. But for public-seceor choices," his impotence with regard
to outcomes implies that there will typically be little effort expended to
learn about government progams. As Downs (1957) concludes: 'Ignorance
of politics is not a result of unpatriotic apathy; rather it is a highly rational
response to "the faces of political life in a large democracy' (p. 147).
- - - ---
On the Nature and Measurement of Fiscal Illusion
67
Imperfect information is not, however, synonymous with fiscal illusion.
It is a necessary, but not a sufficient, condition for its existence. More
specifically, fiscal illusion refers to a systematic misperception of fiscal
a recurring propensity, for example, to underestimate one's
parameters
tax liability associated with certain public programs. Imperfect information
alone might well give rise to a random pattern of over- and underestimation
of such tax liabilities. Fiscal illusion, in contrast, implies persistent .and
consistent behaviour. As such, it will give rise to recurring, and presumably
predictable, biases in budgetary decisions.
It is by no means clear in which direction fiscal illusion will tend to
bias public-sectOr outcomes. In the earlier literature on this matter, Downs
(1960) and Galbraith (1958) argued that imperfect information would tend
to make the public budget too small. It was Downs's contention that the
benefits of most government programs tended to be 'remote' and largely
unrecognised by the electorate, while the taxes to support these programs
are more directly experienced and understood.2 The more pronounced
tendency towards a systematic underestimation of public-seCtor benefits
than costs would lead the electorate to support an unduly small allocation
of resources to the government sector.
However, in the more recent public choice literature, the view has shifted
radically. The attention to special interest groups and associated lobbying
efforts has called into question the presumed lack of support for public
spending. At the same time, studies of revenue structure and 'tax consciousness' suggest that significant elements of the tax system are largely
hidden and 'underperceived' by taxpayers. From this perspective, it is the
costs of government programs, not their benefits, that are subject to significant and regular underestimation. This may stem, in part, from deliberate
efforts by public agencies (of the sort Niskanen envisioned) to disguise
the full costs of their programs and, where possible, to exaggerate the
associated benefits. The tax system, in consequence, has come to include
important elements like tax withhol4ing and forms of taxation with obscure
patterns of incidence that conceal the real cost of public programs (Brennan
and Buchanan, 1980). Fiscal illusion, under this view, results in a public
sector of excessive size.
Not surprisingly, the recent empirical literature on fiscal illusion has
focused exclusively on the revenue side of the budget; it consists of studies
of tax and debt illusion. These studies search for evidence linking relatively
hidden elements in the revenue structUre to higher levels of taxes or
spending.
Before turning to this body of research, there is one general property
of fiscal illusion that is worthy of note and that has some implications
for' empirical testing. Fiscal illusion (or at least tax illusion), by its very
nature, .can only operate over a limited range. Certain kinds of taxes may
remain hidden while they absorb relatively modest fraCtions of personal
-
.
u..
_. _. .. _ _.....
Taxation and Fiscal Federalism
68
income. But as they come to constitute larger and larger chunks of income,
they become progressively harder to hide. To take a specific example, it
has been argued that more income-elastic revenue systems are conducive
to more rapid growth in the public sector, because they generate larger
'automatic' increases in public revenues as the economy grows (Oates, 1975).
if the tax system were
However, there are obviously limits to this
sufficiently income elastic, the argument would imply that the public sector
could come to absorb virtually all of the taxpayer's income with little
opposition. Fiscal illusion obviously cannot persist to such extremes; its
potential is limited to some range of tax bills. Empirical work on tax
illusion may, in consequence, need to make provision for various types
of 'threshhold' effects or, more generally, for varying magnitudes of illusion
over different values of the tax parameters.
-
Empirical Studies of Fiscal Illusion: The General Approach
The bulk of the empirical work on fiscal illusion consists of econometric
studies based upon some kind of model of public spending. The models
range from ad hoc expenditure equations to more rigorously founded demand
functions for public goods (e.g., Bergstrom and Goodman, 1973). In both
cases, these studies end up regressing a measure of budgetary size (either
or, in some instances, changes in spending or
spending or revenues
revenues) on a set of explanatory variables including income, often some
kind of price term, and a series of 'taste' variables. In the Bergstrom and
Goodman formulation, for example, this set of explanatory variables represents the determinants of the demand of the median voter for local public
outputs. These are the variables that would determine public outputs in
the absence of any fiscal illusion.
The next step is to add to the equation variables that will capture any
illusion effects. As we shall see in the subsequent sections, these variables
take a number of different forms in the attempt to test for the presence
of tax or debt illusion. The typical model is thus of the form:
-
1. E
=
aX
+
/3F + u,
where E is a measure of budgetary size, X is a vector of explanatory variables
in a' world without fiscal illusion, F is a vector of variables designed to
measure various types of fiscal illusion, and u is the usual disturbance term.
The null hypothesis for each form of fiscal illusion is taken to be a zero
coefficient on the relevant variable in the F-vector; since the illusion
hypotheses usually imply particular signs for the variables, the alternative
hypothesis is associated with either a positive or negative sign on /3, and
a one-tail test is employed to accept or reject the null hypothesis.
Case I: Complexity of the Revenue System
The first source of fiscal illusion that we shall examine is the complexity
of the revenue system. Buchanan, incidentally, attributes the revenue-
-
-
-
On the Nature and Measurement of Fiscal Illusion
69
complexity hypothesis to Puviani (1903). As Buchanan puts it 'To the
extent that the total tax load on an individual can be fragmented so that
he confronts numerous small levies rather than a few significant ones,
illusory effects may be created' (1967, p. 135). According to this hypothesis,
the more complicated the revenue system, the more difficult it is for the
and the more
taxpayer to determine the 'tax-price' of public outputs
likely it is that he will underestimate the tax burden associated with public
programs. In short, the hypothesis implies that, other things equal, the
more complex the revenue system, the larger will be the public budget.
Richard Wagner (1976) undertook the first test of the revenue-complexity
hypothesis. Wagner's approach, as discussed in the preceding section, was
to regress total current expenditure for a sample of 50 large US cities
on a set of socio-economic variables and a measure of the complexity of
the revenue system. For the Wagner. study, the vector F in equation (1)
collapses into a single variable, f. Just how one measures the complexity
of a revenue system is far from obvious. Wagner, interestingly, chose an
index, the Herfindahl index, that is widely used in the industrial-organisation
literature to measure the degree of concentration within an industry. More
specifically, Wagner's measure of revenue-complexity is:
-
4
2. f
=
k
i
=
rj
1
where rj is the fraction of total city revenue generated from tax source
i. For the rj, he adopted the Census Bureau classification of city-owned
revenues into four categories: property taxes, general sales taxes, selective
excise taxes, and charges and fees. The Herfindahl index will achieve its
maximum value of unity if a city generates all of its own revenues from
a single source; the minimum possible value would be one-fourth if revenues
were divided equally amoung the four categories. A higher value of the
index is thus associated with a less complex revenue system so that the
revenue-complexity
hypothesis
posits a negative coefficient for the variable
f.
Wagner's choice of the Herfindahl index has been an influential one,
for, as we shall see shortly, every subsequent study of the revenue-complexity
hypothesis has used this index as the measure of the illusion variable.
In his estimated regression equation, Wagner found that the sign of the
coefficient on the revenue-complexity variable was indeed negative and
that he could reject the null hypothesis of non association at a .01 level
of significance. Moreover, the point estimate of the coefficient suggested
a very substantial magnitude for the effect: lower values of the Herfindahl
index.were associated with sizeable larger levels of public expenditure.
However, as Wagner acknowledged, the Herfindahl index has some
serious deficiencies as a measure of tax illusion. In particular, the 'visibility'
of the four classes of revenue is likely to vary greatly: a heavier reliance
on charges and fees, for example, will surely provide a more direct sense
70
Taxation and Fiscal Federalism
of the cost of public outputs than a similar reliance on selective excise
taxation. We might expect a city making extensive use of charges and
fees to generate a lower level of spending than one which made relatively
the same use of selective excise taxes. This absence of a measure of the
visibility of various revenue sources suggests that Wagner's regression
equation may be misspecified, raising the likelihood of biased estimators
of the coefficients.3
Subsequent econometric work on the revenue-complexity hypothesis has
yielded somewhat mixed results. Munley and Greene (1978), using a sample
of cities much like that of Wagner, found that the resultS were quite sensitive
to the specification of the equation. Making what appear to be some sensible
alterations to the form of the regression equation, they found that the
estimated coefficient of the revenue-complexity variable was no longer
significantly different from zero, suggesting that Wagner's results may not
be very robust. In another study exploring public spending for higher
education, Clotfelter (1976) could find no evidence supporting a fiscal
illusion. The estimated coefficient on the Herfindahl index was not statistically significant. Moreover, Clotfelter went a step further by trying
to control for the visibility of different revenue sources. His results run
counter to the illusion hypothesis; a heavier reliance on more visible revenue
sources (direct taxation) was associated with more, not less, spending.
However, three other studies, drawing on samples from quite different
sorts of jurisdictions, have found support for the hypothesis. Using data
from 110 large Swiss municipalities, Pommerehne and Schneider (1978)
estimated a series of multiple-regression equations in which the coefficients
on the Herfindahl index had the hypothesised negative sign and were
consistently significantly different from zero. Interestingly, they found that
this result was much stronger in a subset of municipalities with representative democracy but no public referenda than where direct democracy
or referenda were used. Baker (1983) regressed state and local tax revenues
on the usual sortS of variables and obtained a negative and marginally
significant estimated coefficient on his Herlindahl index. And, finally,
Breeden and Hunter (1985), in a stUdy of levels of tax revenues in 37
large US cities, found a negative and significant relationship betWeen these
revenues and the Herfindahl index.
While the results from the various studies of the revenue-complexity
hypothesis are somewhat mixed, there appears to be considerable evidence,
based on the Herlindahl-index variable, in support of this type of fiscal
illusion. As we have seen, this evidence takes the form of a negative partial
association betWeen some measure of budgetary size and the value of the
Herfindahl index. However, some further reflection suggests another, and
I believe more compelling interpretation of this association. State and local
officials tend to be painfully aware of tax rates in other jurisdictions and
try to resist getting too far out of line with rates elsewhere. Levels of
On the Nature and Measurement of Fiscal Illusion
71
sales taxation in excess of those in neighbouring jurisdictions, for example,
can induce shoppers to avoid local purchases of taxed goods. Likewise,
relatively heavy taxation of business capital is often thought to deflect
business investment and jobs away from the state or local economy. In
consequence, as expenditUres in a jurisdiction rise relative to those in nearby
areas, officials are likely to seek out new sources of revenue to prevent
tax rates on existing revenue bases from becoming excessively high relative
to other jurisdictions. We thus tend to find that jurisdictions with relatively
high levels of spending adopt more diversified revenue systems. The argument here is that the structUre of the tax system is itself an endogenous
variable, and, in particular, that the desired level of public expenditUre
influences the community's choice of a tax structUre. If this is true, the
existing econometric tests of the revenue-complexity hypothesis are subject
to simultaneous-equation bias ~ and must be viewed with serious
reservatlons.
We have twO competing hypotheses here: the revenue-complexity
hypothesis and what I will call the 'revenue-diversification hypothesis'.
To discern between them is a difficult task, for they both imply a negative
association betWeen the level of the budget and the diversity (or complexity)
of the revenue system. What we really need, of course, is a theory of
the determination of the tax struCtUre that would explain community choices
among alternative sources of revenues. However, we are a long way from
such a comprehensive theory.
Let me offer a couple of other suggestions. First, there are now some
standard econometric tests for endogeneity (e.g., Wu, 1983). It would seem
sensible to subject the revenue-complexity variable in the earlier stUdies
to such testing to get some indication of whether or not this is a real
issue. Second, and more fundamentally, it would be helpful to find some
way to get at the revenue-diversification hypothesis empirically. One possibility suggests itself. Consider a period of rapid budgetary growth (like
the decade of the 1960s for state and local governments in the United
States). If we compare the revenue system at the beginning and at the
end of such a period (especially for relatively high-spending jurisdictions),
the revenue-diversity hypothesis would suggest that we should observe
a shift from a more simplified revenue system initially to a more diversified
system at the end of the period. The pressures of budgetary growth would,
from this perspective, lead to the introduction of new sources of revenue.
Such a test is admittedly not wholly free of some potential illusion effects,
but it does seem to me that in a period of general growth (like that associated
with the 'baby boom'), the presumption of a finding of increased revenue
diversity would constitUte support for the revenue-diversification hypothesis. As things stand now, I think that we must conclude that 'the jury
is still out' as concerns the empirical verification of the revenue-complexity
hypothesis.
--
-
----
Taxation and Fiscal Federalism
72
Case II: Renter Illusion
In the course of estimating demand functions for local public goods, a
variety of authors have consistently turned up a very striking result: other
things equal, jurisdiCtions with a relatively large fraction of renters tend
to spend more per capita on local public services (see, for example, Bergstrom
and Goodman, 1973, and Peterson, 1975). This pervasive and intriguing
finding has generally been attributed to a kind of 'renter illusion'. Property
taxes, the primary source of local tax revenues, are levied on owners of
not on tenants. While such taxes may be passed forward
rental dwellings
in the form of higher rents to tenants, the renter-illusion hypothesis
contends that renters do not perceive this to be the case. Illusion, in this
instance, takes the form of a failure on the part of tenants to understand
the link betWeen the level of local spending and the level of rent that
they pay. Renters thus misperceive their tax-price for local public outputs
they believe it to be zero, or at least less than the tcue tax-price and, hence, support levels of local expenditure in excess of their counterparts
who own homes and pay property taxes directly.
This empirical finding appears in multiple-regression studies of the
general form discussed earlier. Referring back to equation (1), the F-vector
in these studies consists of a single variable indicating the proportion of
the residents in the jurisdiction who are renters. This variable has, in
and typically
nearly all such studies, a positive and significant coeffIcient
implies a substantial effect on levels of public spending.
Here again, however, the illusion interpretation of this finding is open
to serious question. In one recent paper, Martinez-Vazquez (1983) suggests
that there is no illusion here at all. Rather, he argues that renters do,
in fact, have a lower tax-price than owner-occupants, because 'renters
consume less housing than do homeowners of the same income level' (p.
244). Since one's property-tax liability is a function of the amount of housing
consumed, it follows that the local tax share of a renter will be less than
his home-owner counterpart.
There are additional reasons for scepticism concerning the illusion view.
There is considerable uncertainty concerning the incidence of property taxes
on rental housing. There are certainly circumstances under which one would
expeCt forward shifting Onto tenants; if the higher revenues are associated
with improved local services, then the tax-expenditure increase should
translate into a higher demand for rental housing in the jurisdiCtion that
will drive up rents in a world of mobile consumers. Where tax differentials
do not reflect service differentials, in contrast, the burden of the tax may
remain on the owner. Moreover, even if forward shifting occurs ultimately,
.it may take a considerable period of time. The functioning of rental housing
markets with leases for tenants may introduce substantial time lags into
the process of tax shifting. In some ongoing research, Ellen Roche (1985)
contends that these lags are, indeed, significant, and that their effect is
-
-
-
-------
On the Nature and Measurement of Fiscal Illusion
73
to reduce the present discounted value of any increase in tax liability to
tenants. The extreme case, of course, is that of rent control where, depending
on the specific ordinances, owners may be limited in the extent to which
they can pass along increases in property taxes into higher rents.
What all this suggests is that renter support for expanded budgets may
not be based on fiscal illusion at all. Instead, it may well reflect quite
rational behaviour: under local property taxation, renters may, in fact, have
significantly lower tax-prices than do owner-occupants. If true, incidentally,
this phenomenon has some troubling implications, for it suggests that
renters may to some extent be getting a 'free ride' at the expense of homeowners. Perhaps voting on local budgetary measures (as suggested by
Sonstelie and Pormey [1978]) ought to be limited to those who own
property?
Case III: Income
Elasticity
of the Revenue
System
The third instance of fiscal illusion that has been examined empirically
involves the income elasticity of the revenue system. The revenue-elasticity
hypothesis has been stated succinctly by Buchanan (1967), who argues that:
'In a period of rapidly increasing national product, that tax institUtion
characterized by the highest [income] elasticity will tend, other things equal,
to generate the largest volume of public spending' (p. 65).
On first inspection, this proposition appears to imply some rather peculiar
behaviour on the part of taxpayers. As I have put it earlier:
We would expect that individual demands for public services would be based
upon tastes, levels of income, and the cost of these services. With a given costsharing scheme (or set of tax shares), we could then determine the pattern
of individualdemands at some moment in time.
But why should people care about the income elasticiryof the tax struCtUre?
What the proposition under study seems to imply is that people will not object
to increases in public expenditure if they can be funded with no increases in
tax. rates (that is, from increments to revenues resulting solely from growth
in income), but they will not support an expanded public budget if it requires
a rise in tax rates. This suggests what people care about is not their tax bill,
but rather their tax rate. Viewedthis way,the hypothesis simply is not consistent
with our conventional descriptionof rational behaviour;it implies that consumertaxpayers are subject to a kind of 'fiscalillusion' [Oates, 1975,p. 141].
However, in the context of existing fiscal institUtions, the hypothesis IS
surely plausible (Goetz 1977). As Richard Wagner has argued:
There also exists a large body of casual evidence supporting the existence of
faulry fiscal perception. Whenever a legislative assembly considers changes in
tax rates, tortuous discussion takes place and considerable publiciry is given
to the deliberations. Yet no similar agonizing takes place over the continual,
automatic increase in tax rates that is produced by progressiviry in the national
-----
74
Taxation and Fiscal Federalism
tax structure. Everyoneis aware of a consciouslyenacted tax surcharge; a similar
surcharge is enacted each year when income grows under progressive taxation,
but many taxpayers remain unconsciousof this surcharge [po87].
The issue at this point becomes an empirical one: do revenue systems
with a relatively high income elasticity tend, during periods of economic
growth, ro be associated with more rapid budgetary expansion? The initial
empirical test of this issue (Oates, 1975) examined the growth in spending,
first, by state governments in the US, and, second, by a sample of US
ciry governments over the decade 1960 ro 1970, a period of unusually rapid
budgetary growth in the state and local sector. The srudy followed the
general approach outlined earlier except that the dependent variables were
changes in spending (over the decade) and the independent variables
likewise rook the form of changes instead of absolute levels. The mulripleregression equations thus seek to explain the growth in spending as a
function of changes in the levels of the independent variables and of a
measure of the income elasticity of the revenue system. In the cross-sectional
equations both for state governments and for a sample of 33 large cities,
I found a positive and statistically significanr partial association of tax
elasticity with expendirure growth, providing support for the revenueelasticity hypothesis. However, the magnitude of the effect, as calculated
from the point estimates, was, in my judgmenr, modest.
Subsequenr empirical work has produced mixed results. In a study of
US state government spending, Craig and Heins (1980) found a positive
and significant relationship between tax elasticity and the !eve! of expenditure. They claim, incidentally, that because of 'bounds' on spending, the
appropriate dependenr variable is the level of public spending, not the
growth (or change) in expendirure. This is a tricky matter. As I suggested
earlier, there are indeed limits to illusion effects. Fiscal illusion is likely
ro operate over certain ranges of budgetary variables, but all changes in
these variables surely cannot go unnoriced. Thus, it is not entirely clear
juSt how illusion relationships should be specified. My conjecture is that
neither simple growth nor level measures are fully adequate; we probably
need to think in terms of threshold effects, or, more generally, in terms
of effects that vary in magnirude with budgetary levels. At any rate, Craig
and Heins did find support for the revenue-elasticity hypothesis in terms
of expendirure levels of state governments.
In conrrast, Dilorenzo (1982) found a negative and significant relationship berween his measures of tax elasticiry and expendirure growth
for a sample of 66 US county governments. Dilorenzo attributes this to
Tiebour-like migration effects in response to fiscal differentials, which he
. contends could offset any illusion effects at the local level. However,
it
is hard ro see how this could result in a negative effect! Other studies
include Baker (1983), who finds only marginally significant effects of his
tax-elasticiry variables in explaining levels of tax revenues in the state-
---
--
On the Nature and Measurement of Fiscal Illusion
75
local sector, and Breeden and Hunter (1985), whose results on this issue
I find a little hard to interpret.
Once again the evidence in support of the hypothesis is somewhat mixed.
But there are certainly some results suggesting that tax elasticity matters.
One problem that plagues all of these studies is measuring the income
elasticity of the revenue system. This is not a simple matter where the
system consists of a diverse set of tax bases with varying rates. In the
United States, the Advisory Commission on Intergovernmental Relations
(1968, Table 7; 1974, Table 173) has published some information on tax
elasticity. But the ACIR estimates are seriously incomplete and typically
apply only to a single year. Craig and Heins used the ACIR estimates,
but Oates tried to develop some other proxy measures involving the extent
of reliance on income taxation. There is a real need in this work for improved
measures of revenue elasticity.
.
Ignoring these measurement problems, suppose that we accept the findings of Oates and Craig-Heins of a positive and significant association
of expenditure growth (or level) with revenue elasticity. As mentioned
earlier, there is still the troublesome issue of the endogeneity of the tax
structure. Jurisdictions that prefer relatively high levels of spending may,
for example, opt for more elastic revenue systems. The elasticity variable
may itself be endogenous.
There is a further matter of interpretation that I find even more troublesome. Suppose that we were to conclude that the evidence supports
che revenue-elasticity hypothesis. Can we infer from this che presence of
fiscal illusion? I chink noc. There is another explanation for this phenomenon
chat I find ac leasc as compelling as che illusion view: che transactions
cOltI inherent in modifying budgetary- parameters. Wagner (1971) argues
chat:
As viewed by legislators, the cost of thus changing tax structures may exceed
the benefits, in which case the required change in relative tax collections will
not occur. Such legislative behavior may be interpreted as a form of habitUal
behavior, and may very well be rational. There is a cost incurred in failing
to routinize many types of activity
the value of the resources consumed
in examining the consequences of the alternative choices, in this case, choices
among possible tax structUres. If the costs incurred by continual reexamination
exceed the benefits received from taking more timely actions, habitUal behavior
is efficient
it is the least inefficient of the attainable alternatives [pp. 8586].
-
-
Note thac the transactions-cost explanation does nOt rely on imperfect
voter information or misperceptions. It simply recognises that there are
costs associated with the legislative actions needed to raise tax rates. Such
costs may impede desired rate increases, whereas with a more elastic revenue
system the increase in 'effective' rates will be forthcoming automatically.
The illusion and transaCtions-cost explanations for the revenue-elasticity
...
76
. . -- ...
.. --. -.- -.. ... -.. -. .. - - .- -- - -.. .-- -
---.---
Taxation and Fiscal Federalism
phenomenon are certainly not mutually exclusive. But we cannot, I think,
make an unequivocal leap from the verification of the revenue-elasticity
hypothesis to the conclusion that we have discovered a form of illusion
in the fiscal system.
Case IV: Debt Illusion
A further potential kind of fiscal illusion involves the choice between debt
and tax finance. The argument here is that individuals are more likely
to perceive the costS of public programs if they pay for them through
current taxation than if tax liabilities are deferred through public-sector
borrowing. Vickrey (1961), for example, has referred to '... "a public debt
illusion" under which individuals pay no attention to their share in the
liability represented by the public debt..: (p. 133). Reliance on debt, rather
than tax finance should, from this perspective, tend to result in a larger
public budget.
For rational and informed taxpayers, the choice between these tWOforms
of finance (assuming a similar pattern of incidence) should make no difference. There is obviously an equivalence (Buchanan, 1964 calls it the
'Ricardian equivalence') between a current tax payment and the present
discounted value of the future tax liabilities in the event of debt finance.4
There is now a large literature on the nature of public debt and its effects
on decisions (e.g., see Barro, 1974), which I shall not try to summarise.
But the central proposition here is a straightforward one: there is a fiscal
illusion present when taxpayers tend to underestimate the present discounted value of their future tax liabilities under bond finance. And with
imperfect knowledge, the failure to recognise such furore tax liabilities,
at least to their full sum, seems a real possibility.~
The determination of the existence and the extent of such a debt illusion
is an empirical issue. Once again, this is not an easy matter to get at
empirically. There is, however, a straightforward approach in terms of
local finance. As Daly (1969) and Oates (1972) have pointed out, the furore
tax liabilities associated with local public debt should be capitalised into
local property values. Consider two otherwise identical communities that
undertake identical capital projects. One community, call it A, finances
the project Out of current revenues, while community B chooses to employ
bond finance so as to spread out the payments for the project over future
years. At the end of the current year, the sole differences between a 'typical
resident' of A and B will be that the latter will have a furore tax liability
whose present discounted value equals the recent differential tax payment
by the resident of A. In a world of mobile consumers, this future tax
liability associated with residing in community B will become capitalised
into lower property values in B.6
This suggests an empirical test of the debt-illusion hypothesis. Other
things equal, we should find, if there is a debt illusion, that the furore
..-
--. -..
--.
---
On the Nature and Measurement of Fiscal Illusion
77
tax liabilities associated with the debt are not fully capitalised into local
property values. There has, in fact, been such a study by Epple and Schipper
(1981) in which they examined the partial association betWeen differentials
in local property values across a sample of jurisdictions in the Pittsburgh
metropolitan area and the extent of deferred funding for pensions for
municipal-government employees. Epple and Schipper find no evidence
of debt illusion. Their results suggest that unfunded pension obligations
tend to be fully capitalised into local property values. We obviously cannot
place great confidence in the findings from a single stUdy, but, as yet,
there is not to my knowledge any solid, systematic evidence supporting
the debt-illusion hypothesis.
Case V: The Flypaper Effe<:t
Our final example of fiscal illusion involves what is known in the public
finance literature as the 'flypaper effect'. This is a phenomenon that has
been regularly observed in econometric stUdies of inter-governmental
grants-in-aid. More specifically, these stUdies have found that there is a
significantly higher propensity for recipients to increase public expenditUre
in response to lump-sum inter-governmental grants than in response to
equivalent increases in private income.
In principle, as Bradford and Oates (1971) have shown, there is no
reason for rational and informed individuals in a particular jurisdiction
to regard increases in income from grants any differently from increases
in income from other sources. From this perspeaive, inter-governmental
grants are simply a 'veil' (hence the 'veil hypothesis') for transfers of income
directly to individuals.
However, existing econometric findings do not support the veil hypothesis. Instead, they indicate that communities direct much larger fractions
(40 to 50 per cent) of revenue from lump-sum inter-governmental grants
into state and local outlays than from private income (roughly 10 per
cent ).
How can we explain the flypaper effect? Courant, Gramlich, and Rubinfeld
(1979) and Oates (1979) have proposed a form of fiscal illusion that can
account for this phenomenon. In their models, budget-maximising public
officials effectively conceal the lump-sum character of the grant revenues.
What the electorate sees is a reduction in tax rates needed to finance local
spending programs, and this reduction is erroneously viewed as a reduction
at the margin in the 'tax-price' of these programs. The budgetary process
thus transforms what is, in truth, a lump-sum inter-governmental grant
into what is perceived by individuals as a reduction in the tax-price of
local public goods. The result is a willingness on the pare of the local
electorate to support higher levels of spending than if they correctly perceived the relevant fiscal parameters.
Such an explanation for the existence of the flypaper effect is plausible.
78
Taxation and Fiscal Federalism
But once again, there are competing explanations. It is by no means necessary
to posit the existence of a fiscal illusion to rationalise such behaviour.
Romer and Rosenthal (1979), for example, have shown that where public
officials set the budgetary agenda, excessive public spending can result.
In their model, the public agency takes advantage of a relatively unattraCtive
alternative, the 'reversion level' of the budget which obtains if the eleCtorate
rejects the agency's proposed budget. The proposed budget is larger than
that most preferred by the representative voter but still emerges viCtorious
relatively
because it is preferred by the electorate to the alternative
unappealing reversion level of public output. Note that this explanation
does not presume any lack of information or 'illusion' on the part of taxpayer
voters. Moreover, as Romer and Rosenthal (1980) show, their model
generates a different budgetary response to an increase in income than
to an equivalent increase in lump-sum grants. The model can give rise
to a flypaper effeCt (although other outcomes are also possible).
The basic point is simply that it doesn't require any sort of fiscal illusion
to generate a flypaper result. Plausible forms of local budgetary processes
can do the trick. There is no real evidence that I know supporting the
illusion explanation of the flypaper effect. It is one of the possibilities,
but it needs empirical support.
-
Conclusion
In this chapter, we have examined five distinct forms that fiscal illusion
might take, each of which has been explored in the public-finance literatUre.7
Although forms of illusion are surely possible on the expenditUre side
of the budget, all five cases involve instances of revenue illusion. They
are thought to be cases where individual taxpayers are likely to underestimate
their true tax-prices and consequently to support excessively large levels
of public outlays. It is my contention that although all five cases entail
plausible illusion hypotheses, none of them has very compelling empirical
support. In this regard, I wish to offer tWo concluding observations.
First, the general grounds on which I have expressed reservations about
the existing econometric findings are two: endogeneity of the illusion
variables and alternative explanations of the resultS. These are such
commonplace objeCtions to econometric stUdies that one might regard them
as a 'cheap shot'. It is typically not hard to find some grounds for questioning
the exogenous natUre of an independent variable. Moreover, as we all know,
statistical associations do not demonstrate causation
there is always
room for another hypothesis. While this is true enough, the issue is really
one of judgment and additional evidence. How compelling is the other
explanation for the observed result? How likely and how serious is the
potential endogeneity of the suspect variable? It is my sense, as I have
tried to indicate in the paper, that these are probably serious matters in
the fiscal-illusion literarure. The likelihood of endogeneity in certain cases
-
--
--
- --
On the Nature and Measurement of Fiscal Illusion
79
like the revenue-complexity hypothesis seems to me quite high. Further,
the alternative explanations of a number of the findings strike me to be
at least as persuasive as the illusion rationale. In short, I think that there
really are sources of serious reservation that suggest the need for further,
carefully designed empirical testing both to deal with the endogeneity issue
and to distinguish among the various hypotheses.
My second observation concerns the theory of fiscal illusion. This survey
has focused on empirical studies of illusion hypothesis for a good reason:
there is very little underlying theory. The fiscal-illusion literature has
identified a number of instances of potential illusion, but it has not explored
with much care or rigour their conceptual underpinnings. To take a case
in point, consider the revenue-complexity hypothesis. The assertion here
is that as the revenue system becomes more 'complex' (meaning that it
encompasses a more diverse set of revenue sources), individuals exhibit
an increasing tendency to underestimate their true tax-prices. But how
does this occur? One could imagine a very fragmented revenue system
in which each revenue source was tied closely to a particular government
activity. Such a system could provide quite accurate price signals to taxpayer
voters on the costs of various public programs. There is also the important
distinction between marginal tax-price and total tax liability (see Carter,
1982). What is presumably important is the individual's perception of his
tax-price at the margin which may be fairly clear even though some illusion
may be present as to total tax liability. Although the public sector employs
numerous revenue sources, if it is clear where the marginal funds come
from, tax-price may be readily evident (see Flowers, 1977). My point here
is simply that the study of fiscal illusion requires a careful statement of
the way in which tax-price is perceiv~d and how this perception depends
on revenue structure. The existing literature is suggestive and points to
some important and tantalising hypotheses, but the supporting theoretical
and empirical work remains, for the most part, to be done.
Nates
1. Since my concern in this paper is with fiscal illusion, I will not address a major
source of potential illusion operating through the monetary sector: money creation
and inflationary finance as an alternative to tax or bond finance.
2. Galbraith argued along somewhat different lines. It is his contention that extensive
advertising aaivities in the private sector introduce a 'bias' in resource allocation
away from publicly provided services toward private goods. On the potential
for illusion on the benefit side, see also Musgrave (1981).
3. Robert Schwab has pointed out to me that there may well be a systematic
relationship between the value of the Herfindahl index and the degree of revenue
visibility. A high value of the index (indicating heavy reliance on a single revenue
source) might well be correlated positively with the fraaion of revenues from
-.""
80
4.
5.
6.
7.
-
Taxation and Fiscal Federalism
property taxation. If the property tax is a relatively visible source of revenues,
then the Herfindahl index might be serving as a proxy for visibility as well
as a measure of revenue simplicity.
There are, of course, circumstances that will distUrb this equivalence, as, for
example, where public-sector borrowing can be undertaken at an interest rate
other than that available to the individual tax-payer in the private market. But
the basic point remains. For a careful examination of the Ricardian equivalence
theorem, see Brennan and Buchanan (1980).
According to Buchanan (1967, pp. 132-3), Puviani (1903) had a somewhat more
subtle view of debt illusion. While acknowledging the Ricardian equivalence
of debt and tax finance, Puviani contended that taxpayers would still have a
preference for debt finance, because they would retain 'control' over a captial
value (even though it would be offset by a corresponding liability). The preference
for bond finance based on this control consideration is called by Buchanan an
'asset illusion'.
This is a particularly straightforward case of debt finance: the issue of bonds
to finance a capital project. Debt finance can, however, take other forms. One
increasingly prevalent mode involves 'borrowing' from current public employees
in the form of relatively low wages with compensation in the form of generous
pensions to be paid upon retirement. Because of favourable tax treatment, this
is an attraCtive way for local governments to compensate their employees;
'underfunding' of public pension programs may be a rational course of action
for a community. On this, see Mumy (1978).
There are a few other cases of fiscal illusion that have received isolated study.
Dilorenzo (1982), for example, has found that where municipal utilities use
their profits to subsidise local spending, budgetary levels tend to be higher than
in the absence of such subsidisation.
References
Advisory Commission on Inter-governmental Relations, Federal-Slate-Local Finances: Significanl Features of Fiscal Federalism Washington, D.C.: ACIR., Feb.,
1974.
Advisory Commission on Inter-governmental Relations, Sources of Increased Slale
Tax CoJJeaion: Economic Growlh vs. Polilical Choice, Report M-41 Washington,
D.C.: AOR, Oct., 1968.
.
Baker, Samuel H., 'The Determinants of Median Voter Tax Liability: An Empirical
Test of the Fiscal Illusion Hypothesis', Public Finance Quarterly 11 Jan., 1983,
95-108.
Barro, Robert]., 'Are Government Bonds Net Wealth?', Journal of Polilical Economy
82 Nov.-Dec., 1974, 1095-1117.
Bergstrom, Theodore c., and Goodman, Robert P., 'Private Demands for Public
Goods', American Economic Review 63 June, 1973,280-96.
Bradford, David, and Oates, Wallace, 'The Analysis of Revenue Sharing in a New
Approach to Collective Fiscal Decisions', Quarterly Journal of Economics 85
April, 1971 416-39.
On the Nature and Measurement of Fiscal Illusion
81
Breeden, Charles H., and Hunter, William J., 'Tax Revenue and Tax StruCtUre',
Public Finance Quarterly 13 April, 1985,216-24.
Brennan, Geoffrey, and Buchanan,James M., 'The Logic of the Ricardian Equivalence
Theorem', Finanzarchiv 38 No.1, 1980,4-16.
Brennan, Geoffrey, and Buchanan, James M., The Power to Tax: Analytical Foundation! of a Fiscal Constitution Cambridge: Cambridge University Press, 1980.
Buchanan, James M., 'Public Finance in Democratic Proceu' Chapel Hill: University
of North Carolina Press, 1967.
Carter, Richard, 'Beliefs and Errors in Voting Choices: A Restarement of rhe Theory
of Fiscal Illusion', Public Choice 39 1982,343-60,
Clotfelrer, Charles T., 'Public Spending for Higher Education: An Empirical Tesr
of Two Hypotheses', Public Finance 31 No.2, 1976, 177-95.
Courant, Paul, Gramlich, Edward, and Rubinfeld, Daniel, 'The Stimularive Effecrs
of Intergovernmental Grants: Or Why Money Sticks Where It Hits', in P.
Mieszkowski and W. Oakland, eds" Fiscal Federalism and Grants-in-Aid Washingron, D.c.: The Urban Insrimre, 1979, 5-2l.
Craig, Eleanor D" and Heins, A. James, 'The effect of Tax Elasticity on Government
Spending', Public Choice 35 No, 3, 1980267-75.
Daly, George G., 'The Burden of the Debt and Furure Generations in Local Finance',
Southern Economic Journal 36 July, 1969,44-51.
DiLorenzo, Thomas J., 'Tax Elasricity and the Growth of Local Public Expendirure',
Public Finance Quarterly 10 July, 1982,385-92.
Dilorenzo, Thomas J" 'Utility Profirs, Fiscal Illusion, and Local Public Expendimres',
Public Choice 38 No, 3, 1982,243-52.
Downs, Anrhony, An Economic Theory of Democracy New York: Harper and
Row, 1957.
Downs, Anthony, 'An Economic Theory of Polirical Acrion in a Democracy', Journal
of Political Economy 65 April, 1957, 135-50.
Downs, Anthony, 'Why rhe Government Budget is Too Small in a Democracy',
World Politics July, 1960.
Epple, Dennis, and Schipper, Katherine, 'Municipal Funding: A Theory and Some
Evidence', Public Choice 37 1981, 141-78.
Flowers, Marilyn R., 'Multiple Tax Sources, Voting Equilibrium, and Budgerary
Size', Public Finance 32 No.2, 1977,210-24.
Galbrairh, John K., The Affluent Society Boston: Houghton-Mifflin, 1958.
Goetz, Charles J., 'Fiscal Illusion in Srate and Local Finance', in T, Borcherding,
ed., Budgets and Bureaucrats: The Sources of Government Growth Durham,
North Carolina: Duke University Press, 1977, 176-87.
Martinez-Vazquez, Jorge, 'Renters' illusion or Savvy?', Public Finance Quarterly
11 April, 1983,237-47.
Mumy, Gene E., 'The Economics of Local Government Pensions and Pension
Funding', Journal of Political Economy 86 June, 1978, 517-28.
Munley, Vincent G., and Greene, Kenneth V., 'Fiscal Illusion, the Namre of Public
Goods and Equation Specification', Public Choice 33 No.1, 1978,95-100.
Musgrave, Richard A., 'Leviarhan Cometh - Or Does He?', in H. Ladd and T.N,
Tideman, eels., Tax and Expenditure Limitation! Washington, DC: The Urban
Insrimte, 1981, 77-120.
Oates, Wallace E., Fiscal Federalism New York: Harcourt Brace Jovanovich, 1972.
_--- .- ----_.....-
82
Taxation and Fiscal Federalism
-
Oates, Wallace E., 'Automatic Increases in Tax Revenues
The Effea on the
Size of the Public Budget', in W. Oates, ed., Financing the New Federalism:
Revenue Sharing, Conditional Grants, and Taxation Baltimore: Johns Hopkins
University Press, 1975, 129-60.
Oates, Wallace E., 'Lump-Sum Intergovernmental Grants Have Price Effeas', in
P. Mieszkowski and W. Oakland, eds., Fiscal Federalism and Grants-in-Aid
Washington DC: The Urban Institute, 1979,23-30.
Peterson, George E., 'Voter Demand for Public School Expenditures', in J. Jackson,
ed., Public Needs and Private Behavior in Metropolitan Areas Cambridge, Mass.:
Ballinger, 1975,99-120.
Pommerehne, Werner W., and Schneider, Friedrich, 'Fiscal Illusion, Political InstitUtions, and Local Public Spending', Kyklos 31 No.3, 1978,381-408.
Puviani, Amilcare, Teoria del/a 1//usione Finanziaria Milan: Remo Sandon, 1903.
Roche, Ellen, 'The Incidence of Property Taxes on Renters', unpublished paper
1985.
Romer, Thomas, and Rosenthal, Howard, 'Bureaucrats vs. Voters: On the Political
Economy of Resource Allocation by Direct Democracy', Quarterly journal of
Economics 93 Nov., 1979,562-87.
Romer, Thomas, and Rosenthal, Howard, 'An InstitUtional Theory of the Effect
of Intergovernmental Grants', National Tax journal 33 December, 1980, 4518.
Sonstelie, Jon C, and Portney, Paul R., 'Profit Maximizing Communities and The
Theory of Local Public ExpenditUre', journal of Urban Economics 5 April, 1978,
263-77.
Vickrey, William, 'The Burden of the Public Debt: Comment', American Economic
Review 51 March, 1961, 132-37,
Wagner, Richard E., The Fiscal Organization of American Federalism Chicago:
Markham, 1971.
Wagner, Richard E., 'Revenue StructUre, Fiscal Illusion, and Budgetary Choice',
Public Choice 25 Spring, 1976,45-61.
Wu, D.M., 'Alternative Tests of Independence berween Stochastic Regressors and
Disturbances', Econometrica 41 July, 1973, 733-50.